What it measures and what it doesn’t measure
The severity index, or GI, relates the time off work associated with recorded accidents to the hours worked by the group in question. It is one of the accident rate indices and provides a different dimension than simply counting accidents. Two units with the same number of incidents may have different severity indices if the durations of sick leave differ.
The name might suggest it encapsulates the full severity of an accident, but its scope is more limited. The suffering, the long-term consequences, or the potential impact of a situation are not described by a single ratio. A high-potential incident without lost time can reveal a critical risk even if it doesn’t increase this indicator. Therefore, it must be interpreted in conjunction with qualitative information.
Formula and unit of measurement
The standard formula is: IG = days of sick leave × 1,000 / hours worked. The result is expressed as sick leave days per thousand hours. The numerator and denominator must correspond to the same time frame, with an explicit time rule. The multiplier is one thousand; mistakenly using the one million from the frequency index changes the result by a factor of one thousand.
Internal reports should show all three elements: days, hours, and index. This allows you to identify whether a variation stems from longer absences, more incidents, or a reduction in hours worked. Presenting only the ratio makes it more difficult to review the accuracy of the calculation and can mask significant changes in the volume or composition of activity.
Which days should be counted
NTP 1211, in accordance with the Spanish statistical methodology it cites, includes the calendar days of temporary incapacity during the initial period, including the dates of sick leave and return to work, and excluding subsequent relapses from this calculation. This definition should be distinguished from simply counting shifts or working days not performed. Combining both criteria prevents a proper comparison of the data series.
It is also necessary to define how to handle absences that extend beyond the end of the period. Indicators of days elapsed during the month and final duration of absences due to accidents occurring that month follow different rules. The organization must document the criteria, identify provisional data, and apply the same methodology when comparing with an external source.
Practical example
In a hypothetical example with finalized data, the included accidents result in 120 days of sick leave, and the group has 200,000 hours worked. The calculation is 120 × 1,000 / 200,000 = 0.6. The result is presented as 0.6 days of sick leave for every thousand hours worked. It does not represent 0.6 percent or an individual injury probability.
If another unit accumulates 40 days with the same 200,000 hours, its severity index (IG) is 0.2. The difference may guide a review, but it does not, on its own, demonstrate that the second unit has better control of all risks. Before concluding, specific accidents, tasks, time periods, recording criteria, and potential serious scenarios that did not result in lost work time must be examined.
Ongoing sick leave and revision of results
When a leave of absence remains open, its final duration is not yet known. The report must indicate whether it uses accumulated days up to a cut-off date or waits until cases are closed according to its methodology. A provisional value may increase later without a new accident having occurred, simply because the available information about a previous event has changed.
To avoid confusion, it’s advisable to keep track of the extraction date and the calculation version. If reviewing data from previous periods, the chart should display the updated series with a methodological note. Comparing a completed fiscal year with one that still contains open losses can underestimate the consequences of the latter and create an impression of improvement that will disappear when the data is consolidated.
Differences with frequency and incidence
The frequency index relates accidents to hours worked, while the incidence index uses the working population as a reference. The severity index incorporates the duration of absences. Reading them together allows us to distinguish, for example, a situation with fewer accidents but longer absences from one with more incidents but shorter durations.
This should not be confused with the average duration of sick leave, which divides the number of days by the number of accidents included. Each formula has its denominator and its unit. Creating a table with a name, definition, data source, and update frequency reduces errors when transferring indicators between reports, spreadsheets, or dashboards used by different departments.
How to use it to decide on measures
A change in the IG should prompt an analysis of the underlying causes and a review of working conditions. If several prolonged absences are linked to a single operation, it may be necessary to review its risk assessment and controls. Preventive measures should consider the potential severity and the obligation to act, without relying solely on the magnitude of the indicator’s change.
It is advisable to combine the results with leading and lagging indicators that allow for verification of the implementation of measures. Management should prioritize reliable data and adequate attention to people. Pressuring to reduce absences or alter records in order to improve a figure distorts the information and detracts from the objective: preventing harm and improving actual working conditions.
Common mistakes and limitations
Common errors include including days lost by groups other than those represented in the denominator, mixing calendar and working days, or comparing provisional and final data. Changing criteria without recalculating the series can also distort the result. Historical systems or those of other countries may use different conventions, so the coincidence of the name “severity index” does not guarantee equivalence.
A low IG does not guarantee the absence of non-compliance, nor does it eliminate the need to control serious risks. In small teams, a single case can significantly influence the outcome. Providing context, units, and limitations makes the indicator useful for learning; using it as an isolated ranking can lead to decisions unrelated to effective prevention.
